Three Essays On The Economics Of Health Behaviors

120  Download (0)

Full text

(1)

THREE ESSAYS ON THE ECONOMICS OF HEALTH BEHAVIORS

A Dissertation

Presented to the Faculty of the Graduate School of Cornell University

in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy

by Kai-Wen Cheng

(2)
(3)

THREE ESSAYS ON THE ECONOMICS OF HEALTH BEHAVIORS

Kai-Wen Cheng, Ph.D. Cornell University 2010

This dissertation investigates the economics of health behaviors. It focuses on the ways health behaviors, specifically smoking and fertility, respond to economic factors such as price and income, as well as non-economic factors such as health-related knowledge and health policy. The first chapter, “The effect of contraceptive knowledge on fertility: the roles of mass media and social networks,” explores the effect of contraceptive knowledge on fertility using an instrumental variables approach. It draws upon the “Knowledge, Attitudes, and Practice of Contraception in Taiwan” (KAP) dataset and focuses on the period when Taiwanese family planning programs were in effect. The results indicate that mass media and social networks play important roles in disseminating contraceptive knowledge. This study finds that women transform their knowledge into behavior--that is, contraceptive knowledge reduces fertility, no matter which fertility metric is measured (life-time fertility or probability of giving birth). The second chapter (coauthored with Donald Kenkel), “U.S. cigarette demand: 1944 – 2004,” uses data from 23 national cross-sectional surveys conducted by the Gallup Poll from 1944 through 2004 to investigate the changes in cigarette demand in the United States from the 1940s through 2004, individual and government attitudes toward smoking changed dramatically. It estimate standard two-part models of cigarette demand as a function of demographics, income, and cigarette prices. The results show that from 1944 to 2004: the gender difference in smoking rates almost disappears; the black-white difference reverses; a strong gradient with schooling emerges; and the negative income elasticities

(4)

strengthened in magnitude. The third chapter, “WTO Entry, a New Cigarette Tax Scheme, and the Tobacco Market in Taiwan,” analyzes the impacts of Taiwan’s entry into the WTO, which was accompanied by a series of policy changes on both the supply and demand sides of the tobacco market. It investigates the link between cigarette tax and price by imputing the tax pass-through rates, and confirms the

hypothesis that free trade induces an increase in advertisements and the introduction of new brands and products. Regarding smokers’ reactions to price changes, this study finds some evidence that smokers not only react to price changes, but also react to relative price changes by switching brands. It also takes into account other scenarios accompanying the WTO entry that influence the brand choices.

(5)

BIOGRAPHICAL SKETCH

Kai-Wen Cheng was born in Keelung, Taiwan on November 19, 1978. She obtained her bachelor and master degrees in Economics at National Taipei University, Taipei, Taiwan in 2000 and 2002, respectively. She then worked as a research assistant in the department of Economics at National Taiwan University, Taipei, Taiwan. In 2004, she joined the Ph. D program in the department of Policy Analysis and Management at Cornell University. In the summer of 2009, she defended her dissertation. She then joins the post-doctoral fellowship program in the Center for Tobacco Control Research and Education at University of California at San Francisco.

(6)

To my parents: Chin-Huang Cheng(鄭金煌) and Chiung-Chu Huang(黃瓊珠) for all their sacrifices, support, and love.

(7)

ACKNOWLEDGEMENTS

I would like to express the deepest appreciation to my advisor, Donald Kenkel, for his supervising during my five-year in the Ph. D program, and his knowledge sharing for carrying out the research projects. He sets a good example of being an unbiased researcher and person -- I learned a lot from him academically and

non-academically. Without his guidance and persistent encouragement this

dissertation would not have been possible. I would also like to thank my committee members, John Cawley and Daniel Lichter for their comments and suggestions to improve this dissertation. I also thank Dean Lillard for his comments and insights on my research projects.

I thank my field friends, Julie Carmalt, Daniel Wilmoth, Hua Wang, Fenaba Addo, Ning Zhang, Feng Liu, Asia Sikora, and Mabel Andalon for being such amazing friends in supporting me academically and non-academically. I also thank my housemates, Holly Yang, Vicky Lo, Lillian Hsu, and Pei-Yu Liao for their emotional support for the life away from the home country.

I am grateful to my parents and sister, for their unconditional love and sacrifices. I thank my Dad for always taking his time out calling me and talking with me, no matter how busy he is. I thank Sheng-Wei Chi for always being there for me during the long research process.

Finally, I would like to acknowledge the generous funding support received from the R01 HD048828 from the National Institutes of Health, the Hu Shih Memorial Award from East Asian Program at Cornell University, Cornell Graduate School, and the Department of Policy Analysis and Management.

(8)

TABLE OF CONTENTS

Biographical Sketch... iii

Dedication... iv

Acknowledgements ... v

Table of Contents... vi

List of Figures... viii

List of Tables ... ix

Introduction ... 1

Chapter 1: The Effect of Contraceptive Knowledge on Fertility: the Roles of Mass Media and Social Networks Introduction ... 3

Taiwan’s Family Planning Programs ... 7

Literature Review ... 9

Data... 11

Identification Strategy ... 14

Results ... 17

Evaluating the IV Strategy... 28

Robustness Check... 30

Conclusion/Discussion ... 32

References ... 34

Chapter 2: U.S. Cigarette Demand: 1944-2004 Introduction ... 38

Literature Review ... 41

Data... 48

Reliability of Gallup Poll Data ... 49

Descriptive Summary ... 50

Empirical Models and Variables... 53

Results ... 56

Regressions by Decades ... 56

Regressions of Pooling Years ... 66

The Cohort Effects: Pooling Years from 1944 to 2004 ... 66

The Influences of Economic Factors Pooling Years from 1970 to 2004... 69

Omitted Variable Bias in Price Elasticity ... 73

Discussion... 75

References ... 77

Chapter 3: WTO Entry, a New Cigarette Tax Scheme, and the Tobacco Market in Taiwan Introduction ... 81

Background of the Tobacco Market and New Cigarette Tax Scheme ... 83

The Economics of 2001 Policy Changes... 86

Data... 89

The Relationship between Taxes and Prices... 90

New Brands, Nicotine, and Advertisements before and after 2002... 95

Impacts on Cigarette Consumption ... 97

Smoking Prevalence ... 97

(9)

Brand Switching Behaviors between 2000 and 2001... 101

Other Scenarios Influence the Brand Choices... 104

Conclusion ... 104

References ... 106

(10)

LIST OF FIGURES

Chapter 1: The Effect of Contraceptive Knowledge on Fertility: the Roles of Mass Media and Social Networks

Figure 1: The Prevalence of Contraceptive Knowledge 1965, 1967, 1976, 1980, 1985 ... 14 Chapter 2: U.S. Cigarette Demand: 1944-2004

Figure 2: Smoking Participation and Cigarette Prices, 1954-2004 ... 75 Chapter 3: WTO Entry, New Cigarette Tax Scheme, and the Tobacco Market in Taiwan

Figure 3: Cigarette Prices between Domestic and Imported Cigarettes 2001 and 2002 ... 86

(11)

LIST OF TABLES

Chapter 1: The Effect of Contraceptive Knowledge on Fertility: the Roles of Mass Media and Social Networks

Table 1-1: Summary Statistics ... 12

Table 1-2: The Number of Live Births, OLS Model ... 18

Table 1-3: The Number of Live Births, 2SLS Model (first stage)... 20

Table 1-4: The Number of Live Births, 2SLS Model (second stage)... 22

Table 1-5: The Probability of Giving Births, OLS Model... 23

Table 1-6: The Probability of Giving Births, 2SLS Model (first stage) ... 25

Table 1-7: The Probability of Giving Births, 2SLS Model (second stage) ... 27

Table 1-8-1: The Number of Live Births Control for Family Attitudes ... 30

Table 1-8-2: The Probability of Giving Births Control for Family Attitudes... 31

Chapter 2: U.S. Cigarette Demand: 1944-2004 Table 2-1: Comparing the Smoking Status Reported in NHIS and Gallup Poll... 50

Table 2-2: Descriptive Statistics ... 52

Table 2-3-1: Smoking Participation (Linear Probability Model with State Fixed Effects) ... 57

Table 2-3-2: Smoking Participation (Linear Probability Model with Clean Indoor Air Index)... 58

Table 2-3-3: Smoking Participation (Linear Probability Model) ... 58

Table 2-3-4: Smoking Participation (Linear Probability Model with State Fixed Effects and Clean Indoor Air index) ... 59

Table 2-4-1: Smoking Level (OLS Model with State Fixed Effects) ... 60

Table 2-4-2: Smoking Level (OLS Model with Clean Indoor Air Index) ... 61

Table 2-4-3: Smoking Level (OLS Model) ... 61

Table 2-4-4: Smoking Level (OLS Model with State Fixed Effects and Clean Indoor Air Index) ... 62

Table 2-5: The Cohort Effects: Pooling Years from 1944-2004 ... 67

Table 2-6-1: Smoking Participation: Pooling Years from 1970 to 2004 ... 69

Table 2-6-2: Smoking Level: Pooling Years from 1970 to 2004... 70

Table 2-7: Imputed Price Elasticity and Income Elasticity from Two Part Model 1970 – 2004 ... 72

Table 2-8: State Level Models of Cigarette Prices ... 74

Chapter 3: WTO Entry, New Cigarette Tax Scheme, and the Tobacco Market in Taiwan Table 3-1: Cigarette Tax Policy Before and After the New Tax Scheme in 2002 .. 85

Table 3-2: Cigarette Prices and Taxes across Brands before and after 2002... 91

Table 3-3: Cigarettes Magazine Advertisements 2001-2002... 96

Table 3-4: The Transition of Smoking Status in 2001 and 2002 ... 97

Table 3-5: Brand Switching between 2001 and 2002... 99

(12)

Introduction

Health behaviors or lifestyle factors have tended for decades to be regarded as the major determinants of premature mortality. Healthy behaviors directly improve the overall public health of a nation while reducing the cost of health care systems. Public and private agencies have, accordingly, conducted interventions such as increasing taxes, providing health-risk knowledge, implementing restrictions, etc. as attempts to alter citizens’ unwanted behaviors and encourage their healthy behaviors.

This dissertation investigates the economics of health behaviors. It focuses on the ways health behaviors, specifically smoking and fertility, respond to economic factors such as price and income, as well as non-economic factors such as

health-related knowledge and health policy. It also examines health behaviors across different socio-economic statuses.

The first chapter, “The effect of contraceptive knowledge on fertility: the roles of mass media and social networks,” investigates the ways people respond to new health-related knowledge. It focuses on the period when Taiwan’s family planning programs were in effect and examines the relationship between contraceptive knowledge and fertility. The implementation of family planning programs is an example of providing information intended to change behaviors: information about contraceptive techniques is provided to women of childbearing age so that they will increase the practice of contraception and thus control fertility. The chapter

examines how individuals build their contraceptive knowledge from these programs; how socioeconomic characteristics, mass media exposure, and an individual's social network influence the forming of that contraceptive knowledge; and whether the obtained contraceptive knowledge reduces fertility.

The second chapter and third chapter focus smoking behavior in the U.S. and Taiwan. The second chapter, “U.S. cigarette demand: 1944 – 2004,” investigates the

(13)

changes in cigarette demand in the United States from the 1940s through 2004, individual and government attitudes toward smoking changed dramatically. It examines the changing influences of smoking determinants on smoking behaviors over time. Because the data cover a long time span, we are able to study cigarette demand before and during the early years of tobacco control efforts, as well as during the most recent period.

The third chapter is “WTO entry, new cigarette tax scheme, and the tobacco market in Taiwan”. This study analyzes the impacts of Taiwan’s entry into the WTO in 2002 and the accompanying policy changes, on both the supply (cigarettes

producers) and demand sides (cigarettes consumers) of the market. In particular, it studies the tax pass-through rates for each individual brand and product. It

investigates the non-price competition of tobacco companies reflected in new

introduced products and brands, and marketing promotion as well. Finally, it studies smokers’ reactions to price changes in terms of brand switching.

(14)

CHAPTER 1: THE EFFECT OF CONTRACEPTIVE KNOWLEDGE ON FERTERTILITY: THE ROLES OF MASS MEDIA AND SOCIAL NETWORKS

Abstract

This study explores the effect of contraceptive knowledge on fertility using an instrumental variables approach. It draws upon the “Knowledge, Attitudes, and Practice of Contraception in Taiwan” (KAP) dataset and focuses on the period when Taiwanese family planning programs were in effect. This study differs from previous studies examining the effectiveness of family planning programs on fertility by focusing on individuals’ obtained contraceptive knowledge and fertility. The results indicate that mass media and social networks play important roles in disseminating contraceptive knowledge. Women who are regularly exposed to mass media, or who have a wider social network, have more knowledge about contraceptives than their counterparts. This study finds that women transform their knowledge into

behavior--that is, contraceptive knowledge reduces fertility, no matter which fertility metric is measured (life-time fertility or probability of giving birth). Since very few studies focus on the relationship between contraceptive knowledge and fertility, by exploring this relationship, this paper contributes to an improved understanding of how the individuals obtain the disseminated knowledge; how socioeconomic characteristics, mass media exposure, and social network influence the forming of knowledge; and whether the obtained knowledge is transformed into new behaviors.

Introduction

Many advertising campaigns sponsored by private or public agencies disseminate health, nutrition, and product information aimed at changing people’s behaviors. Such information about issues reaches its goal only if individuals obtain

(15)

the disseminated information and transform the acquired information into new behaviors. This study focuses on the period when Taiwan’s family planning programs were in effect and examines the relationship between contraceptive knowledge and fertility. It examines how the individuals build their contraceptive knowledge from the programs; how socioeconomic characteristics, mass media exposure, and social network influence the forming of that contraceptive knowledge; and whether the obtained contraceptive knowledge reduces fertility.

The implementation of family planning programs is an example of providing information intended to change behaviors: information about contraceptive techniques is provided to women of childbearing age so that they will increase the practice of contraception and thus control fertility. The ultimate aim is to couple low birth rates with a consistently low mortality rate to reduce population growth. For developing countries where the population transitions from a combination of high mortality rate and high birth rate to a combination of low mortality rate and high birth rate, the resulting rapid population growth may create pressures on housing, education, and social patterns. Such a situation often dramatically increases the financial burden of the nation as a whole. In order to control population growth by reducing fertility rates, governments may opt to implement family planning programs which provide married couples with information about modern contraceptive techniques,

contraceptive access, and the benefits of having fewer children. In some societies, such programs may also aim to overcome entrenched gender preference toward sons.

Several studies have focused on investigating whether such family planning programs play any role in decreasing fertility, or whether the decrease is actually driven by economic and social changes; for example, improved educational and economic opportunities for women might cause them to desire fewer children.1

1

(16)

However, the endogenous characteristics of the input-allocation of family planning programs – high fertility villages tend to be the target of family planning programs and hence receive more family planning inputs than other areas – make the evaluation of the causal effect of family planning programs challenging2.

This study differs from previous studies examining the effectiveness of family planning programs on fertility by focusing on individuals’ obtained contraceptive knowledge and fertility. This study examines the factors related to the acquisition of contraceptive knowledge, and the relationship between an individual’s contraceptive knowledge and their fertility during the period when family planning programs were enacted. Since dissemination of information relating to modern contraceptive techniques is one of the main ways for family planning programs to control fertility, examining the ways married women obtain contraceptive knowledge from the programs; the differences in knowledge acquisition across different demographic, social, and economic clines; and the subsequent effects on fertility sheds new light on the effectiveness of family planning programs, as well as the relationship between contraceptive knowledge and fertility.

Taiwan’s family planning programs, enacted nationwide in 1964, aimed to decrease the fertility rate in order to control population growth. To reach this goal, the programs educated citizens about population growth issues, extolled the benefits of smaller families, and valuing daughters as highly as sons. These aim at changing married couples’ traditional values about family. In addition, the program provided information about accessing and using contraceptive techniques. In Taiwan in the 1960s, primary education was not universal and both public transportation and

2

There are a few of randomly designed family planning programs, such as the Taichung city experiment conducted in 1963, the Matlab family planning program, and the family planning programs

(17)

communication technologies were limited;3 the family planning program therefore used a variety of information dispersal techniques, including visiting families, placing advertisements/announcements in mass media, and encouraging word-of-mouth communication via friends, relatives, or neighbors to disseminate the information on family planning programs. For example, the information about modern contraceptive techniques, modern contraceptive access, and the benefits of having fewer children. Previous literature (for example, Kan and Tsai, 2004; Aggarwal and Rous, 2006; Barber and Axinn, 2004; Montgomery and Casterline, 1993; Behrman et al., 2002) have found that mass media exposure and word-of-mouth communication play

important roles in obtaining the disseminated information in developing countries such as Taiwan, Nepal, India, and Kenya.

The detailed information on women’s contraceptive knowledge, fertility history, mass media exposure, women’s organization participation, and household and

demographic characteristics in the “Knowledge, Attitude and Practice of

Contraception in Taiwan” data sets allow researchers to measure directly women’s contraceptive knowledge; contraceptive knowledge across socioeconomic

characteristics, mass media exposure, and social networks; and the outcomes on fertility.

However, the obtained contraceptive knowledge is jointly determined by factors related to the demand- and supply-side of contraceptive knowledge.

Unobserved factors, such as a couple’s modernization and their sex/ parity/ quantity preference toward children, determine the levels of demand for both fertility and contraceptive knowledge. The existence of unobserved factors makes identification of causality challenging. This study uses an instrumental variables approach to

3

In 1964, around 22% of the population did not have primary education, 3.9 per thousand households had the motor transportations, and 11.6 per thousand households had a telephone set in Taiwan. (Source: Taiwan Statistical Data Book)

(18)

resolve the endogeneity issue. Mass media exposure and connection to social networks are treated as instrumental variables of contraceptive knowledge to examine the causal effect of contraceptive knowledge on fertility.

There have been several studies investigating the relationship between knowledge and behaviors applied to different fields of interest, such as product consumption, risky behaviors, and health outcomes. Very few studies, however, focus on the relationship between contraceptive knowledge and fertility4. By exploring this relationship, this paper contributes to an improved understanding of the relationship between knowledge and behavior.

Taiwan’s Family Planning Programs

Taiwan’s death rate fell from about 14 to 5 per thousand between 1948 and 1962, while the fertility rate remained unchanged. High fertility rates and low death rates led to an annual rate of population growth that reached 3.5% in the years between 1951 and 1956. The 3.5% growth rate caused the population to double in only 20 years (Freedman and Takeshita, 1969). Although it is possible for social and economic development to change the role of the traditional family and decrease the demand for children, it usually takes years to complete the transition from high mortality and fertility to low mortality and fertility. Therefore, Taiwan’s family planning programs were implemented nationwide to slow down population growth and shorten the period of demographic transition to prevent a large population growth that might impede economic development.

Taiwan’s family planning programs were enacted nationwide in 1964. Before 1964, there were some voluntary and quasi-governmental activities advocating family

4

Goldin and Katz (2002), Bailey(2006), Ananat et al (2007) use access to, rather than knowledge of, contraceptive techniques to analyze its effect on age of first marriage, professional career, and life-time

(19)

planning. For example, in 1950 the Joint Commission for Rural Reconstruction (JCRR) issued one million copies of the pamphlet, “The Happy Family,” advocating family planning by the rhythm method. In 1954, the China Family Planning

Association, a voluntary organization subsidized by the JCRR, organized a training program emphasizing birth control and child spacing for women living in the dependent villages (Freedman et al., 1994).

Around 1963 and 1964, there was an experimental study in the city of

Taichung to test the effectiveness of a more intensive family planning program. This study established that many families were interested in family planning and that couples in all social strata would accept contraceptive techniques when they were offered. The success of the program provided support for a later nationwide family intervention. In 1964, the government started a nationwide five-year plan, with a grant of US $24 million, to reduce the fertility rate by persuading 600,000 women to use contraceptives for their family planning needs.

The program involved 300 female health workers who made motivational and educational visits to women of childbearing age in their homes to offer subsidized contraceptives (Freedman et al., 1994). Since the number of pre-pregnancy health workers was limited, they concentrated first on visiting families with more than three children, those with sons, those living in high-fertility counties, the poor, and those living in remote villages. The reason was that these women had a stronger

motivation to accept contraception, and their higher acceptance rates would most effectively lower the overall fertility rate.

The family planning program also used public media, such as radio, TV, newspapers, and slides at Taiwan’s movie houses to explain contraceptive techniques and how to obtain contraceptives. Articles on family planning were clipped out every month from 15 of Taiwan’s 22 newspapers. In 1965 there were a total of 319

(20)

articles related to family planning (Chu, 1966). In addition, around 50,000 posters were printed and placed in villages around the island. Mass media and word-of-mouth communication are the main ways to disseminate the contraceptive information. Over 60% of married women indicate they obtained the information about family planning from mass media or friends/ relatives/neighbors5.

The government also used financial incentives to encourage women to use contraception. When new kinds of contraceptive techniques were introduced, the government updated their method of subsidizing contraceptives. The government first encouraged using loop and subsidized half of the cost; then they started to encourage using contraceptive pills and condoms and subsidized part of the cost. In addition to the government’s subsidization of sterilization surgery for the poor, each city government also used welfare funding to subsidize sterilization surgery for the general population (Freedman et al., 1994). The number of people undergoing surgical sterilization rose rapidly. The family planning programs were officially ended in 1985.

Literature Review

There have been several studies investigating the relationship between knowledge and behaviors that have focused on different fields of interest, such as consumption and health-related behaviors. Some studies measure an individual’s information acquisition about issues and examine the individual's subsequent behavior according to different information acquisition (for example, Kenkel, 1991; Kan and Tsai, 2004; Nayga, 2000); others focus on an event shock, such as the removal of the ban on nutrition claims on product and advertising style campaigns, to identify the information effect to examine different reactions among different subgroups toward

(21)

the new information (for example, de Walque, 2004; Ippolito and Mathios, 1999). Regarding the literature about factors related to fertility, there is a large body of literature, covering several different countries, investigating the relationship between contraception and fertility. Most of this literature focuses on contraceptive access rather than knowledge. Several studies focus on family planning programs in the developing countries. They use the time and location variation among family planning programs as inputs to investigate the effect of contraception access on

fertility (for example, Miller, 2008). A few studies focus on fertility in the U.S.; they use abortion legalization and pharmaceutical regulations, which vary states and over time, to examine the effect of contraception accessibility on fertility related outcomes (Goldin and Katz, 2002; Bailey 2006, Ananat et al, 2007)). These studies

demonstrate that women who have access to contraception at an early age have fewer births and better career achievement than those without such access.

In addition, a large body of literature has focused on the individual's decision to use (or not to use) contraception and their choice of contraception types focusing on institutional and social factors influencing the decisions. Institutional factors shape the accessibility and availability of contraceptives which directly influence the use and choice of contraception (Braunder-Otto et al. 2007). Social effects, on the other hand, influence contraceptive adoption through defining it as a social acceptable behavior, and by spreading the information and adoption of new behaviors (Montgomery and Casterline, 1993; Behrman et al., 2002; Edmeades, 2008). Institutional effects and social effects may jointly influence the adoption of new behaviors. Institutional effects may indirectly influence the new behavior by establishing a social and economic environment which relates to the diffusion and adoption of new behaviors (Edmeades, 2008).

(22)

similar to the process of decision making about contraceptives. The institutional effects and social effects influence the dissemination of contraceptive knowledge in the same way that they influence contraceptive practice and choice. The mass media campaigns/ advertisements sponsored by family planning programs could be seen as an institutional effect because they indirectly influence women’s awareness of modern contraceptives, not only by spreading information about contraceptive methods, but also by identifying locations for obtaining contraceptives. These campaigns can also be seen as social effects, as they shape contraception as a social acceptable behavior. Social networks, through which the contraceptive knowledge spread, are another method by which a social effect influences the establishment of contraceptive knowledge and multiplies the effect of mass media on the build of contraceptive knowledge. Several studies focus on factors such as mass media and social networks, associating them with the establishment of health-related information (Kan and Tsai (2004); Aggarwal and Rous (2006); Barber and Axinn (2004); Montgomery and Casterline (1993); Behrman et al. (2002)).

This study adds to the existing literature by examining the effect of

contraceptive knowledge obtained from several mechanisms, such as mass media and social networks, on fertility.

Data

This research is primarily based on data from five island-wide surveys, “Knowledge, Attitudes, and Practice of Contraception in Taiwan” (KAP). They are repeated cross-sectional data conducted respectively in 1965, 1967, 1976, 1980, and 1985.6 These surveys interviewed married women of reproductive age (18-44). The

6

I do not include KAP 3 data collected in 1970, because the nature of KAP 3 is different from the other sets of KAP. KAP 3 re-interviewed half of the respondents interviewed in 1967, while the other half of

(23)

data set includes information about women’s fertility history, desired number of children, and attitudes toward, knowledge of, and use of contraception. In addition, measures of socio-economic status and demographic information such as age,

education, employment, and family history for both wives and husbands are covered. Table 1-1 presents the summary statistics for each survey year. In 1965, the year after the nation wide implementation of family planning programs, the women in fertile ages have had 4.04 live births on average. In 1967, two years after, the

average number of live births drops to 3.96 and it keeps dropping to 2.66 live births in 1985. On the other hand, contraceptive knowledge among married women of fertile age is expanding over time. In 1965, married women know about 3.5 modern contraceptive techniques on average; in 1967, married women know about 4 modern contraceptive techniques, and in 1980s, married women know about 8 modern contraceptive techniques. Figure 1 presents the prevalence of knowledge of the selected modern contraceptive techniques for married women in every survey year. It shows that the prevalence of each specific technique might reflect the target of contraception that family planning programs emphasize. For example, the family planning programs first encouraged practicing loop, ota ring, and tubal ligation; later on, the programs encouraged women to use condoms and oral pills. The data in Figure 1 is consistent with that pattern. Furthermore, the practices of contraception and abortion have been increasing (Table 1-1). In 1965, only 27% of married women ever practiced contraception; however, in 1985, 88% of them ever practiced

contraception. In 1965, only 10% of married women had ever had an abortion; in 1985, 28% of them had had one or more abortion.

Table 1-1: Summary Statistics

KAP 1 (1965) KAP 2 (1967) KAP 4 (1976) KAP 5 (1980) KAP 6 (1985) Sample size 3,719 4,989 5,587 3,852 3,819

(24)

Table 1-1 (continued)

Dependent variable

Number of living births 4.04 3.96 3.20 2.70 2.66

Ideal number of births 3.96 3.89 3.25 2.84 2.57

Abortion 0.10 0.12 0.20 0.23 0.28

Number of abortion 1.57 1.60 1.57 1.60 1.54

Number of contraceptive techniques known 3.5 4.00 6.15 8.05 7.97

Contraception practice 0.27 0.41 0.68 0.83 0.88

Independent variable

Whether the respondent reads the newspaper often

0.14 0.21 0.29 0.51 0.62

Whether the respondent reads magazines often 0.11 0.16 0.22

Whether the respondent listens to the radio often

0.54 0.15 0.27 0.49 Whether the respondent watches the TV often 0.21 0.70 0.79 0.93 Does the respondent own a Radio 0.55

Whether living with other married couples 0.27 0.28 0.16 0.10 0.11

Organization participation 0.15 0.06 0.06

No son 0.18 0.15 0.18 0.20 0.18

Women’s Education level

Illiterate 0.49 0.40 0.25 0.12 0.08

Elementary 0.41 0.49 0.59 0.58 0.48

Junior high 0.06 0.06 0.08 0.11 0.17

Senior high 0.03 0.04 0.05 0.13 0.21

College 0.01 0.01 0.02 0.06 0.06

Husband’s years of schooling 5.78 6.19 7.32 8.60 9.26

Whether working outside of family 0.17 0.20 0.44 0.31 0.33

Whether living with parents or parent’s in law 0.52 0.46 0.39 0.38 0.40

Women’s age 31.95 32.15 33.49 30.70 32.20

Living in city dummy 0.30 0.31 0.43 0.47 0.50

Sample size 3,719 4,989 5,587 3,852 4,312

Husband’s ethnicity (Fukiennese) 0.76 0.73 0.68 0.72 0.73

(Hakka) 0.14 0.13 0.12 0.12 0.14

(Mainlander) 0.09 0.14 0.17 0.12 0.09

Contraceptive Knowledge

Know Condom 0.29 0.30 0.54 0.85 0.89

Know Foam Tablets 0.29 0.28 0.29 0.36 0.24

Know Jelly 0.17 0.15 0.23 0.35 0.26

Know Diaphragm 0.12 0.14 0.19 0.35 0.38

Know Rhythm 0.20 0.27 0.45 0.60 0.67

Know Basic Temperature 0.05 0.09 0.20 0.43 0.53

Know Coitus Interruption 0.04 0.08 0.24 0.45 0.50

Know Ota Ring 0.64 0.61 0.76 0.87 0.75

Know Loop 0.47 0.62 0.89 0.96 0.92

Know Oral Pill 0.31 0.47 0.85 0.93 0.93

Know Vasectomy 0.33 0.35 0.65 0.93 0.93

(25)

the prevalence of contraceptive

knowledge

0 0.1 0.2 0.3 0.7 0.9 1 967 1 2 4 9 0.4 0.5 0.6 0.8 percentage 3 5 6 7 8

# of

contraceptions

Know Condom Know Oral Pill Know Ota Ring Know Loop Know Tubal 1965 1 1976 1980 1985 year 0 Ligation number of known

Figure 1: The Prevalence of Contraceptive Knowledge 1965, 1967, 1976, 1980, 1985

The increasing trends of mass media exposure, women’s education levels, urban residence, and women’s working status also reflect the rapid social chang economic development of Taiwan during the 1960s-1980s. More and more wom were regularly exposed to radio, TV, newspapers, and magazines over time. Women’s education levels and working status also increased.

Identification Strategy The direction of causati

es and en

on between contraceptive knowledge and fertility

behavio ;

hoosing to rs is a concern. One possibility is that contraception choices affect fertility women who have larger contraceptive knowledge are more resourceful in c

among different kinds of contraceptive techniques and practice the contraception control their fertility. Another possibility is that fertility affects the acquisition of contraceptive knowledge. Women who have reached their desired number of

(26)

children, or have achieved their desired gender ratio among their children, have incentives to seek out more contraceptive knowledge than those who have not. Finally, external factors may determine levels of both fertility and contraceptive knowledge. For example, women who are more “modern” and “westernized” are mo open to and resourceful with modern contraceptive techniques, and they at the sa time demand fewer children.

Therefore, the Ordinary Least Squares (OLS) model without correcting the endogeneity in contraceptive knowledge, does not gauge the true effect of

contraceptive knowled

re me

ge on fertility. This study uses an instrumental variables pproach to overcome the endogeneity issue, using mass media exposure and women’s participation in organizations as th ntraceptive knowledge. The

a h

y

tion. Section 7 explains in detail the strength and validity of these ins

ssue, a

e instruments of co

hypotheses are: 1) married women who regularly listen to the radio, watch TV, read magazines, or read newspapers have more access to contraceptive advertisements and family planning campaigns, and hence, obtain more contraceptive information; 2) married women who actively participate in community-based organizations have wider social network, and hence, obtain more contraceptive knowledge throug word-of-mouth communications. The instruments are believed to influence fertilit only through contraceptive knowledge; that is, they are uncorrelated with the error term in fertility equa

truments.

Life-time Fertility

First, I used the OLS model, which does not take into account endogeneity i to estimate the life-time fertility equation (1) to investigate the relationship between contraceptive knowledge and fertility. N refers to the number of live births by the i woman i; K is the number of contraceptive techniques the woman i has heard of; i

(27)

i

X refers to other variables influencing fertility, such as the woman’s age cohort,

education, husband’s education, husband’s income, husband’s ancestry, her current working status, urban/rural residence, cohabitation with parents-in-law, and other factors. i i i i K X N012 (1)

Second, I take into account the endogeneity of contraceptive knowledge. In order to overcome the endogeneity issue, I use the two-stage least square (2SLS)

pproach: first, I use mass media exposure and organization participation as the instruments to identify the in equation (2), and then I use th

ve

-of-mouth communication). a

effect of contraceptive knowledge

e predicted value of contraceptive knowledge from (2) to estimate the effect of contraceptive knowledge in the fertility equation (3). The variables indicating whether the respondents regularly watch TV, listen to the radio, read newspapers, or read magazines are proxies for exposure to the fertility-related campaigns and contracepti advertisements in the mass media. The variables indicating whether they participate in community organizations are proxies for exposure to contraceptive knowledge through social networks (word

i i i i i i i i IV IV IV IV IV X K01 1 +γ2 2 +γ3 3 +γ4 4 +γ5 5 +γ6 (2) i i i i =β0+β1Kˆ +β2X +ε (3)

The Probability of Giving Birth

The number of live births is recorded from the year of marriage to the current year, while contraceptive knowledge, women’s working status, urban/rural res

and cohabitation with parents-in-law, and the measures of mass media exposure an social network are measured in the current year. To ensure the exa

N

idence, d mination of the causal effect of knowledge on fertility, all variables are measured in the current state. I use linear probability model to examine the likelihood of giving birth in the previous

(28)

year in equation (4)7. ) ( ) Pr(Bt1,i =G β01Kt,i2Nt1,i3Boyt1,i4Xt,ii (4) i t

B1, is a binary variable indicate whether the married women had a live birth last

year; Kt,i is the current contraceptive knowledge; Nt−1,i is the number of live births until the last year; Boyt−1,iis the number of boy births until the last year; Xt,i refers to other variables influencing fertility, such as the woman’s age cohort,

education, husband’s education, husband’s income, husband’s ancestry, her current working status, urban/rural residence, cohabitation with parents-in-law, and other factors. OLS model without correcting the endogeneity issue is first est

In addition, I take into account th neity of contraceptive knowledge and use the 2SLS approach. I first use mass media exposure and organization participation as the instruments to ident fect of contraceptive knowledge in equation (5), and then I use th ptive knowledge from (5) to estimat

ing imated. e endoge

ify the ef

e predicted value of contrace

e the effect of contraceptive knowledge in the fertility equation (6). I used the same set of instrumental variables as the total number of live birth equation indicat whether the respondents regularly exposed to mass media and their connections to social networks. i i i i i i i i IV IV IV IV IV X K01 1 +γ2 2 +γ3 3 +γ4 4 +γ5 5 +γ6 (5) ) ˆ ( ) Pr(Bt1,i =G β01Kt,i2Nt1,i3Boyt1,i4Xt,ii (6) Results

Life-time Fertility Equation

Table 1-2 presents the regression results of the number of live births estimated with OLS. The regressions are estimated separately by each survey year. Wife’s

This study examines the birth probability in the previous year instead of the current year, because it ensures the duration of each possible event occurs is one year and it can be consistent in each survey 7

(29)

education level and current working status, which could serve as proxies for prices having children, are negatively associated with the number of live births. Husband’s education is negatively correlated with number of live births. Husband’s income is not statistically associated with the number of births.

Older women have more live births than younger ones. Husband’s ancestry is associated with the number of live births. Compared with the Fukiennese, the Mainlanders have fewer live births. Women who live in the city have fewer births. The OLS model indicates that contraceptive knowledge is positively associated wit the number of live births in the earlier survey years: 1965, 1967, and 1976. The magnitude of this association decreases with time. The sign of coefficient on

contraceptive knowledge changes to negative in the later survey years: 1980 and 1985. The positive relationship between contraceptive knowledge and life time fertili

of

h

ty contrad ition that contraception prevents unintended births, since it does not

e positive relationship explains a

rce of endog ivin posit

le, w g ber o ren a target

ly planning programs m rce he s, such as

sits from health personnel, tele ont ab on e The decreasin nitud ositiv et effects explains that the ect has been vanishing over tim lanat r the positive

ip is the reverse ca y betw ontra e kno e and lity: ho have had a larg ber of births might have mo entive eek out

echn th to re

liv hs, O

OLS: d ari er hs

icts the intu

take into account the endogeneity issue. Indeed, th

possible sou eneity: there might be a target effect dr g the ive relationship. For examp

fami

omen with a very lar e num f child re the

of and get ore resou s from t program

family vi phone c acts, etc out the c traceptiv

knowledge. g mag e of p e targ

target eff e. Another exp ion fo

relationsh usalit een c ceptiv wledg ferti

women w e num re inc s to s

effective contraceptive t iques on eir own prevent p gnancy.

Table 1-2: The number of e birt LS model

ependant v able: numb of live birt

(30)

Table 1-2 (continued) n Number of contraceptives know 0.07 0.05 0.02 -0.02 -0.02 [0.01]** [0.01]** [0.01]** [0.01]* [0.01]+ No son -1.49 -1.64 -1.15 -0.97 -0.90 [0.10]** [0.09]** [0.06]** [0.06]** [0.05]** age1822 -3.95 -3.75 0.00 -1.97 -1.78 [0.14]** [0.11]** [0.00] [0.11]** [0.08]** age2327 -3.27 -3.13 -2.20 -1.44 -1.31 [0.14]** [0.09]** [0.07]** [0.08]** [0.07]** age2832 -2.06 -2.09 -1.45 -0.80 -0.81 [0.11]** [0.08]** [0.05]** [0.08]** [0.06]** age3337 -0.83 -1.01 -0.64 -0.24 -0.37 [0.10]** [0.09]** [0.04]** [0.08]** [0.05]**

Wife is working outside of family

-0.25 -0.17 -0.15 -0.12 -0.21

[0.07]** [0.07]** [0.04]** [0.04]** [0.04]**

Wife years education 12 and er

ov

-0.61 -0.74 -0.58 -0.47 -0.46

[0.14]** [0.12]** [0.07]** [0.04]** [0.05]**

Wife years education 0-6 0.29 0.36 0.37 0.24 0.03

[0.07]** [0.06]** [0.05]** [0.06]** [0.10]

Husband years education 0-6 0.30 0.12 0.20 0.26 0.21

[0.07]** [0.06]+ [0.06]** [0.06]** [0.12]+

Husband years education 12 and o

-0.42 -0.20 -0.30 -0.28 -0.25 ver

[0.07]** [0.08]* [0.05]** [0.04]** [0.05]**

Husband’s ancestry: hakka 0.00 -0.13 -0.06 -0.09 -0.14

[0.09] [0.09] [0.07] [0.06] [0.06]* Husband’s ancestry: -0.56 -0.43 -0.14 -0.19 -0.33 mainlander [0.09]** [0.09]** [0.06]* [0.05]** [0.07]** Live in a city -0.66 -0.22 -0.20 -0.20 -0.22 [0.16]** [0.10]* [0.07]** [0.07]** [0.05]**

Live with parents-in-law -0.08 -0.02 0.05 0.05 0.03

[0.06] [0.05] [0.04] [0.04] [0.04] Husband’s income -0.04 0.00 -0.04 -0.03 [0.04] [0.01] [0.04] [0.03] Constant 6.11 6.38 4.60 3.76 4.03 [0.13]** [0.22]** [0.06]** [0.10]** [0.11]** Observations 3662 4871 4678 3852 3817 R-squared 0.56 0.53 0.49 0.47 0.46

Robust standard errors in brackets

+ significant at 10%; * significant at 5%; ** significant at 1%

In order to resolve the endogeneity issue, a 2SLS model which takes into account endogeneity is estimated. Table 1-3 and 1-4 present the results. The result of the first

(31)

stage is listed in Table 1-3, and the second stage in Table 1-4. The result in Table 1-3 dicates that women who are regularly exposed to mass media, including watching

newspapers have larger ntraceptive k

tions have greater co ve ge ir n rticip

The instruments explain contraceptive knowledge very well. F

. All F statistics are above 10, su ng the shold

enta ables8 e first icates that m edia

in g c tiv ledg

ith the findings of previous literature.

r of live 2SL sta

stage: o pti ues

in

TV, listening to the radio, reading magazines, or reading

co nowledge than those who do not; women who participate in organiza ntracepti knowled than the on-pa ating

counterparts. The

statistics are 26.92, 65.50, 46.06, 14.60, and 29.75 in 1965, 1967, 1976, 1980, and

1985 respectively rpassi thre of

powerfulness for instrum l vari . Th stage ind ass m and social networks play crucial roles obtainin ontracep e know e, consistent w

Table 1-3: The numbe births,

mber

S (first ge)

2SLS: first dv: nu f contrace ve techniq known

1965 1967 1976 1980 1985 No son -0.89 -0.84 -0.61 -0.00 -0.11 [0.13]** [0.12]** [0.12]** [0.12] [0.10] age1822 -0.66 -0.91 0.00 -0.65 0.11 [0.22]** [0.18]** [0.00] [0.31]* [0.19] age2327 -0.16 -0.34 -0.48 -0.12 0.62 [0.14] [0.13]** [0.14]** [0.20] [0.15]** age2832 0.43 0.24 -0.02 0.04 0.41 [0.10]** [0.12]* [0.12] [0.19] [0.12]** age3337 0.41 0.36 0.28 0.18 0.20 [0.13]** [0.12]** [0.09]** [0.16] [0.11]+

Wife is working outside of family

0.00 0.25 0.16 0.07 0.25

[0.14] [0.11]* [0.13] [0.13] [0.09]*

Wife’s education is 12 or above 1.08 0.86 0.84 0.89 1.03

[0.36]** [0.28]** [0.16]** [0.12]** [0.11]** Wife’s education is 0-6 -1.02 -0.80 -0.66 -0.77 -0.88 [0.12]** [0.11]** [0.11]** [0.18]** [0.23]** Husband’s education is 0-6 -0.31 -0.46 -0.46 -0.45 -0.91 [0.10]** [0.09]** [0.11]** [0.13]** [0.27]**

8 Staiger and Stock (1997) suggest that the instru s co ea tatistic

an 10.

ment set i nsidered w k if the first stage F s is less th

(32)

Table 1-3 (continued) usband’s education is 12 or .28 0.98 H above 1 0.78 0.61 0.66 [0.21]** [0.17]** [0.13]** [0.12]** [0.10]**

Husband’s ancestry: hakka 0.68 0.26 0.95 0.03 0.32

[0.33]* [0.15]+ [0.33]** [0.26] [0.17]+

Husband’s ancestry: mainlander 0.17 0.24 0.33 -0.13 -0.44

[0.19] [0.16] [0.17]+ [0.15] [0.14]**

Living in a city -0.14 0.62 0.54 0.04 0.80

[0.47] [0.18]** [0.37] [0.24] [0.29]**

Living with parents in law -0.18 -0.26 -0.27 -0.01 0.07

[0.09]+ [0.09]** [0.09]** [0.08] [0.08]

Listen to radio 0.36 0.82 0.51 0.21 0.30

[0.10]** [0.09]** [0.13]** [0.11]+ [0.09]**

Read newspapers 1.88 1.59 1.17 0.83 1.07

[0.22]** [0.16]** [0.13]** [0.15]** [0.13]**

Live with married couples 0.06 -0.13 -0.39 -0.08 -0.09

[0.11] [0.09] [0.16]* [0.13] [0.10] Husband’s income 0.49 0.09 0.38 0.24 [0.07]** [0.02]** [0.10]** [0.08]** Watch tv 0.72 0.83 0.29 -0.03 [0.13]** [0.12]** [0.13]* [0.15] Read magazines 0.89 0.84 0.54 [0.14]** [0.13]** [0.08]** Join organizations 0.49 0.42 0.52 [0.19]* [0.18]* [0.16]** Constant 4.47 2.19 1.86 6.27 6.33 [0.18]** [0.19]** [0.14]** [0.28]** [0.22]** F statistics 26.92 65.50 46.06 14.60 29.75 Observations 3662 4868 4678 3852 3819 R-squared 0.38 0.43 0.40 0.31 0.35

Robust standard errors in brackets

+ significant at 10%; * significant at 5%; ** significant at 1%

The results of the secon A king

tive to negative. itio rac ch ow men

eases the total number by 09 18 0 i 967, 76, 1980, and 1985 respe The price ef ferti tio tive--

en with high educa cu or side of the fam

The income th e s p ut hes

d stage are listed in Table 1-4. fter ta into account endogeneity, the signs of the coefficients on contraceptive knowledge change from posi An add nal cont eptive te nique kn n by wo

decr of births 0.16, 0. , 0.14, 0. , and 0.2 n 1965, 1

19 ctively. fect in lity equa n is nega

the wom tion and rrently w king out ily have

(33)

statistical significance in 19 e s o em c f

ore live births.

tha nne era

ber of live births, 2SLS (second stage)

LS: ent var numbe e births

67. Th influence f other d ographi actors on fertility are similar with the findings in OLS. Older cohorts have m

Mainlanders have fewer live births n Fukie se on av ge.

Table 1-4: The num

2S depend iable: r of liv

1965 1967 1976 1980 1985 Number of contraceptive techniques known -0.16 -0.09 -0.14 -0.18 -0.20 [0.05]** [0.03]** [0.03]** [0.03]** [0.03]** No son -1.68 -1.76 -1.24 -0.96 -0.92 [0.09]** [0.07]** [0.06]** [0.05]** [0.05]** age1822 -4.10 -3.86 0.00 -2.06 -1.76 [0.14]** [0.12]** [0.00] [0.10]** [0.09]** age2327 -3.31 -3.16 -2.25 -1.46 -1.20 [0.09]** [0.07]** [0.06]** [0.06]** [0.06]** age2832 -1.98 -2.04 -1.45 -0.79 -0.72 [0.08]** [0.07]** [0.05]** [0.06]** [0.06]** age3337 -0.76 -0.96 -0.58 -0.22 -0.33 [0.08]** [0.07]** [0.05]** [0.06]** [0.05]**

Wife is working outside of .24

family -0 -0.14 -0.13 -0.11 -0.18 [0.07]** [0.06]* [0.05]** [0.04]** [0.04]** W ab ife’s education 12 or ove -0.18 -0.55 -0.32 -0.25 -0.20 [0.19] [0.13]** [0.09]** [0.08]** [0.07]** Wife’s education 0-6 -0.03 0.15 0.20 0.06 -0.22 [0.10] [0.07]* [0.05]** [0.06] [0.08]** Husband’s education 0-6 0.22 0.04 0.11 0.18 0.02 [0.07]** [0.06] [0.06]+ [0.07]** [0.11] Husband’s education 12 or -0.00 0.01 -0.0 above 8 -0.13 -0.08 [0.14] [0.09] [0.06] [0.06]* [0.05]

Husband’s ancestry: hakka 0.15 -0.10 0.09 -0.08 -0.07

[0.10] [0.08] [0.07] [0.06] [0.06] Husban mainlander d’s ancestry: -0.52 -0.36 -0.06 -0.21 -0.41 [0.11]** [0.08]** [0.05] [0.06]** [0.07]** Living in a city -0.67 -0.12 -0.06 -0.17 -0.06 [0.11]** [0.10] [0.07] [0.06]** [0.07]

Live with parents in law -0.12 -0.06 0.01 0.05 0.04

[0.06]+ [0.05] [0.04] [0.04] [0.04]

Husband’s income 0.06 0.03 0.04 0.04

[0.04] [0.01]* [0.04] [0.03]

(34)

Table 1-4 (continued) [0.37]** [0.21]** [0.20]** [0.29]** [0.28]** Observations 3662 4868 4678 3852 3817 R-squared 0.50 0.50 0.43 0.40 0.36 Over-identification test (p-value) 0.4707 0.0826 0.1415 0.0307 0.0099 Standard errors in brackets

+ significant at 10%; * significant at 5%; ** significant at 1%

The Probability of Giving Birth

This study estimates the probability of giving birth using OLS and 2SLS pproach. The result of OLS is listed in Table 1-5. The result indicates that

lihood of giving birth in the ear, conditional upon the e birth has been given before

, except in 1967. Those who had not had any sons before the previous y to birth ear than the counterparts. This is true in ey year. The result implies that marri es’ ce ons is

e younger cohorts have a higher probability of giving birth in the Women who are currently working outside of family

probability o bi th year en’ ion is

rth lan le to have

pared ne

ty ving b OLS l

OL eth rt r

a

contraceptive knowledge has almost no effect on the like

last y accumulative liv the

previous year

year yet are more likel give last y

each surv ed coupl preferen toward s

still existent. Th

previous year than older cohorts.

have a lower f giving rth within e past . Wom s educat not associated with the likelihood of giving bi . Main ders are ss likely births last year com to Fukien se.

Table 1-5: The probabili of gi irths, mode

S: dv: wh er having bi hs last yea

1965 1967 1976 1980 1985

Cumulative not having ns

so

0.09 0.18 0.12 0.08 0.11

[0.02]** [0.02]** [0.02]** [0.02]** [0.02]**

Total live births until last year -0.00 -0.00 -0.04 -0.09 -0.09 [0.00] [0.00] [0.01]** [0.01]** [0.01]** Number of contraceptive techniques known -0.00 -0.01 0.00 0.00 0.00 [0.00] [0.00]* [0.00] [0.00] [0.00]+

(35)

Table 1-5 (continued) age1822 0.41 0.43 0.00 0.20 0.26 [0.04]** [0.04]** [0.00] [0.04]** [0.05]** age2327 0.52 0.49 0.31 0.20 0.26 [0.03]** [0.03]** [0.03]** [0.03]** [0.03]** age2832 0.39 0.33 0.28 0.09 0.13 [0.02]** [0.02]** [0.02]** [0.02]** [0.02]** age3337 0.15 0.13 0.15 -0.00 0.00 [0.02]** [0.02]** [0.01]** [0.02] [0.01]

Wife is working outside of family

-0.10 -0.05 -0.13 -0.09 -0.07

[0.02]** [0.02]** [0.02]** [0.01]** [0.02]**

Wife years education 12 or ove

ab

0.00 -0.02 0.03 -0.01 0.00

[0.04] [0.04] [0.04] [0.02] [0.02]

Wife years education 0-6 -0.00 0.02 0.02 -0.01 0.01

[0.02] [0.02] [0.02] [0.02] [0.02]

Husband years education 0-6

0.05 0.04 0.00 0.03 -0.01

[0.02]* [0.02]* [0.02] [0.02] [0.03]

Husband years education 12 and a

-0.03 -0.01 -0.04 -0.04 -0.01 bove

[0.03] [0.02] [0.02]* [0.02]* [0.02]

Husband ancestry: hakka 0.03 -0.00 -0.01 -0.00 -0.02

[0.03] [0.03] [0.02] [0.02] [0.02] Husband ancestry: -0.07 -0.02 -0.08 -0.08 -0.06 mainlander [0.03]* [0.02] [0.02]** [0.02]** [0.02]** Living in a city -0.06 -0.01 -0.03 -0.03 -0.03 [0.02]* [0.03] [0.02] [0.02] [0.02]+

Live with parents-in-law 0.01 0.01 0.02 0.04 0.04

[0.02] [0.02] [0.01] [0.01]** [0.01]**

Husband’s income -0.02 -0.00 0.02 0.02

[0.01] [0.00] [0.01] [0.01]+

Constant 0.08 0.31 0.27 0.34 0.23

[0.04]+ [0.08]** [0.03]** [0.04]** [0.04]**

Observations 3564 4776 4460 3788 3614

R-squared 0.21 0.24 0.21 0.26 0.33

Robust standard errors in brackets

+ significant at 10%; * significant at 5%; ** significant at 1%

The result of 2SL is n T -6 an Table 1-6 presents The resu ort lt i 1-3 that mass m posure

etworks play nt he ion trac knowledge.

edicting contraceptive knowledge. S analys is listed i able 1 d 1-7.

the first stage. lts supp the resu n Table edia ex and social n importa roles in t acquisit of con eptive The instruments are powerful in pr

(36)

Table 1-6: The probabili ng SL ta

t sta mb chniq wn

ty of givi births, 2 S (first s ge)

2SLS, firs ge, dv: nu er of contraceptive te ues kno

1965 1967 1976 1980 1985

Cumulative having no sons

-0.63 -0.70 -0.60 0.02 -0.15

[0.12]** [0.12]** [0.11]** [0.10] [0.10]

Number of live births until last year

0.19 0.17 0.09 -0.09 -0.08 [0.03]** [0.03]** [0.03]** [0.04]* [0.05] Age1822 0.43 0.08 0.00 -0.90 -0.06 [0.25]+ [0.22] [0.00] [0.31]** [0.26] age2327 0.73 0.43 -0.09 -0.31 0.50 [0.19]** [0.16]** [0.16] [0.20] [0.17]** age2832 0.96 0.69 0.22 -0.06 0.33 [0.13]** [0.13]** [0.13] [0.19] [0.14]* age3337 0.64 0.56 0.38 0.16 0.17 [0.13]** [0.12]** [0.10]** [0.16] [0.12]

Wife is working outside of family -0.00 0.26 0.15 0.06 0.23 [0.14] [0.11]* [0.13] [0.13] [0.09]* Wife’s education is 12 or above 1.17 1.00 0.95 0.83 1.00 [0.37]** [0.27]** [0.15]** [0.13]** [0.11]** Wife’s education is 0-6 -1.16 -0.84 -0.69 -0.75 -0.88 [0.13]** [0.11]** [0.11]** [0.18]** [0.23]** Husband’s education is 0-6 1.36 1.00 0.80 0.59 0.64 [0.21]** [0.17]** [0.13]** [0.12]** [0.10]** Husband’s education is .73 12 or above 0 0.26 0.94 0.02 0.31 [0.32]* [0.15]+ [0.33]** [0.26] [0.16]+ Husband’s ancest hakka ry: 0.27 0.31 0.32 -0.14 -0.47 [0.19] [0.15]* [0.17]+ [0.16] [0.13]** Husband’s ancestry: ainlander .03 0.66 m 0 0.57 0.02 0.78 [0.45] [0.18]** [0.37] [0.24] [0.28]** Live in a city -0.16 -0.24 -0.26 -0.01 0.07 [0.09]+ [0.09]** .09]** [0.08] [0 [0.08]

Live with parents in law 0.41 0.84 0.50 0.21 0.29

[0.10]** [0.09]** [0.13]** [0.11]+ [0.09]**

Listen to the radio 1.91 1.63 1.18 0.82 1.07

[0.21]** [0.16]** [0.13]** [0.15]** [0.13]** Read newspapers 0.10 -0.11 -0.40 -0.09 -0.09 [0.11] [0.09] [0.16]* [0.13] [0.10] Live wi couples th married -0.47 -0.47 -0.43 -0.90

(37)

Table 1-6 (continued) [0.09]** [0.11]** [0.13]** [0.27]** Husband’s income 0.48 0.09 0.38 0.24 [0.07]** [0.02]** [0.10]** [0.08]** Watch TV 0.70 0.85 0.29 -0.03 [0.13]** [0.12]** [0.12]* [0.15] Read magazines 0.90 0.84 0.53 [0.14]** [0.13]** [0.08]** Join Organization 0.49 0.43 0.52 [0.18]* [0.18]* [0.16]** Constant 3.16 1.15 1.46 6.57 6.62 [0.23]** [0.25]** [0.18]** [0.29]** [0.24]** F statistics Observations 3662 4868 4678 3852 3817 R-squared 0.39 0.44 0.41 0.32 0.35

Robust standard errors in bracket

+ significant at 10%; * significant at 5%; ** significant at 1%

Table 1-7 presents the second stage. The results show that, conditional on the number of births until last year, one more contraceptive technique known prevents the likelihood of having births by 0.06, 0.03., 0.04, 0.05 and 0.05 last year in survey years of 1965, 1967, 1976, 1980, and 1985 respectively9. The more births each woman has had, the less likely she is to give birth, and the magnitude of this effect increases over

e. This explains the number of births to each woman has been decreasing over time d any sons remain

othe hi

horts ar ke e b n o s. ho are

e fam ive birth. Wo h

re likely to have given b centl ch mi flect the

e relationship betw h ed n and ge an ess

wh ently h p -law

ore likely to have gi re ut ly significant in

tim

which reflects the decreasing birth rate. Women who have not ha more likely to have an

exists. Younger co

r birth. T s shows that sex preference toward sons still e more li ly to hav irths tha lder one Women w currently working outsid of the ily are less likely to g men wit

higher education are mo irth re y, whi ght re

positiv een hig ucatio young a s. Mainl ders are l likely to give births than Fukiennese. Women o curr live wit arents-in are m ven birth cently, b the coefficients are on

9 The proportion of women who gave birth in the previous year is 0.440, 0.386, 0.363, 0.281, 0.262 in each survey year.

(38)

1980 and 1985. Husband’s income is ly d w ro of rth.

lity g b SLS st

2SLS, second stage, dv: whether having births last year

positive associate ith the p bability giving bi

Table 1-7: The probabi of givin irths, 2 (second age)

1965 1967 1976 1980 1985 Number of contraceptive chniques known te -0.06 -0.03 -0.04 -0.05 -0.05 [0.02]** [0.01]** [0.01]** [0.01]** [0.01]**

Cumulative having no sons 0.05 0.16 0.10 0.08 0.10

[0.03]* [0.02]** [0.02]** [0.02]** [0.02]**

Number of live births until st year la 0.01 -0.00 -0.04 -0.10 -0.09 [0.01] [0.00] [0.01]** [0.01]** [0.01]** age1822 0.44 0.44 0.00 0.14 0.25 [0.05]** [0.04]** [0.00] [0.04]** [0.04]** age2327 0.55 0.50 0.31 0.18 0.28 [0.03]** [0.03]** [0.03]** [0.03]** [0.03]** age2832 0.44 0.34 0.29 0.09 0.15 [0.03]** [0.02]** [0.02]** [0.02]** [0.02]** age3337 0.19 0.15 0.16 0.01 0.01 [0.03]** [0.02]** [0.02]** [0.02] [0.02]

Wife is working out mily side of .10 fa -0 -0.04 -0.12 -0.08 -0.06 [0.02]** [0.02]* [0.02]** [0.02]** [0.01]** Wife’s education is 12 or above 0.13 0.02 0.10 0.06 0.08 [0.06]* [0.04] [0.03]** [0.03]* [0.02]** Wife’s education is 0-6 -0.09 -0.01 -0.03 -0.07 -0.08 [0.03]** [0.02] [0.02] [0.02]** [0.03]** Husband’s education is 0-6 0.02 0.02 -0.02 -0.00 -0.06 [0.02] [0.02] [0.02] [0.03] [0.04] Husband’s education is 12 or above 0.08 0.02 0.02 0.01 0.04 [0.04]* [0.03] [0.02] [0.02] [0.02]*

Husband’s ancestry: hakka 0.07 0.01 0.03 0.00 0.00

[0.03]* [0.02] [0.02] [0.02] [0.02] Husband’s ancestry: mainlander -0.06 -0.01 -0.06 -0.09 -0.09 [0.03]+ [0.02] [0.02]** [0.02]** [0.02]** Live in a city -0.06 0.00 0.01 -0.02 0.02 [0.03]+ [0.03] [0.02] [0.02] [0.02]

Live with parents in law 0.00 0.00 0.01 0.04 0.05

[0.02] [0.01] [0.01] [0.01]** [0.01]**

Husband’s income 0.00 0.00 0.04 0.04

(39)

Table 1-7 (continued) Constant 0.32 0.35 0.36 0.75 0.68 [0.10]** [0.06]** [0.07]** [0.12]** [0.11]** Observations 3564 4773 4460 3788 3614 R-squared 0.12 0.23 0.16 0.16 0.23 Over-identification test(p-value) 0.9775 0.6887 0.1788 0.2236 0.7259 Standard errors in brackets

+ significant at 10%; * significant at 5%; ** significant at 1%

In general, the results indicate that contraceptive knowledge reduces fertility no matter whether that fertility is measured by life-time fertility or the probability of recently having given birth.

Evaluating the IV strategy

The instruments of contraceptive knowledge -- mass media exposure and social networ

tive

es

n

y that ks – have a strong joint influence on the obtainment of contraceptive

knowledge. This study would be able to identify the causal effect of contraceptive knowledge on fertility as long as the exclusion restriction is valid, that is, as long as mass media exposure and social networks affect fertility only through contracep knowledge.

Indeed, the over-identification test suggests that the instruments this study us are valid especially in the likelihood of having births equation. The p-values to over identification tests are listed in Table 1-4 and 1-7. The null hypothesis that no

association between contraceptive knowledge and error term in fertility equation fail to be rejected in the year 1965 and 1976 for total number of live births equation, and every survey year for the likelihood of having birth equation.

However, a number of arguments still can be made to question exclusio restriction. First, mass media exposure and/or organization participation might not only expand contraceptive knowledge but also shape fertility attitudes in a wa

(40)

influences fertility demand. If fertility attitudes changed through mass media exposu olely amily. ight n

of women who are not. hey might be different in observable ways. For example, women who are regularly exposed to mass media and/or hav orks might have a higher level of

educati ight be

n than t

ble re influence the acquisition of contraceptive knowledge, then the coefficient on contraceptive knowledge in the second stage of the 2SLS approach does not s

reflect the effect of contraceptive knowledge on fertility; it also reflects the couple’s attitudes about a desired number of children and/or the sex composition of their f Women who are regularly exposed to mass media or who have a wider social network are more likely to have access to family planning messages on the benefits of having fewer children and access to knowledge of modern contraceptive techniques than women without that exposure. If contraceptive attitudes and knowledge are correlated, the coefficient on contraceptive knowledge in the fertility equation m not only reflect the contraceptive knowledge but also attitudes which lead to over-estimate the effect of contraceptive knowledge.

Another argument concerning the validity of the instrumental variables used in this paper is based upon a hypothetical unobserved characteristic which may

collectively drive contraceptive knowledge, mass media exposure and organizatio participation. Women selected to the group with regular exposure to mass media and/or with wider social networks are different from the group

T

e larger social netw

on, be younger, and be wealthier. On the other hand, it is possible they m different in unobservable ways. For example, women with regular mass media exposure and wider social networks might be more open to new informatio those with less exposure. These observable and unobservable characteristics migh influence the fertility decision. While this study controls for differences in observa characteristics, it does not control for differences in unobservable characteristics. If there is an unobservable difference between the two groups in the case described

(41)

above, the coefficient on contraceptive knowledge might be biased upward a reflect the true effect of contraceptive knowle

nd not dge on fertility.

attitudes toward family planning might lead to the dy takes the tudy uses h hav nform on

y planning eral tra vi .

des toward family planning and general nal in ertility quation in 2SLS aims to resolve ential i d g nsis

e effect on fertility.

he results indicate that the s of total live birt

irths are very robust after taking into account women derni

toward family planning. T idence pr s the c ce to usa contraceptive knowledge t dy finds result isted

r of live b l ily attitudes

ths equation (2SLS)

Robustness Check

In order to take into account the potential factors leading to the biased contraceptive knowledge effects listed in the section 7 -- the unobserved married couple’s modernity and

over-estimated contraceptive knowledge effect on fertility. This stu

advantages of the affluent data sets this s whic e the i ation attitudes toward famil and gen ditional ewpoints It controls for

the attitu traditio viewpo ts in the f

e the pot ssues an et the co tent

contraceptive knowledg

T equation hs and probability of

having b ’s mo ty and

attitudes he ev ovide reden the ca l

effect of his stu . The s are l in Table 1-8-1

and Table 1-8-2.

able 1-8-110: The numbe irths contro for fam T

The number of live bir

1965 1967 1976 1980 1985 Contraceptive knowledge -0.17 -0.10 -0.13 -0.18 -0.21 [0.06]** [0.03]** [0.03]** 0.04]** .03]**[ [0 Tradition 0.02 0.17 0.09 0.12 [0.09] [0.07]* [0.04]* [0.04]** Attitudes 0.42 0.44 0.27 0.06 0.11 [0.11]** [0.08]** [0.09]** [0.08] 4]**[0.0

10 The variable, tradition, measures the married women’s viewpoints toward tradition. Tradition is coded with “1” if the respondents answer “definitely yes” or “probability yes” to the question “do you expect to live with your children or grandchildren in old age?”; and “0” otherwise. The variable, attitude, measures the married women’s viewpoints toward family planning programs. Attitude is coded with “1” if the respondents answer “approve very much” or “approve” to the question “are you in favor of family planning/ contraception?”; and “0” otherwise.

Figure

Updating...

References

Updating...

Related subjects :